M. Todd - you're right in that not all Z is treated equally by the ZQ score, and it's possible I'm being too hard on the ZQ score.  At the same time, I still very much believe that it has limits in terms of usefullness in determining sleep quality from an objective point of view.  When you look at the breakdown of ZQ score for the night, it appears to me that the weight of Total Z is very large compared to the weight of getting lots of deep or REM sleep.  Perhaps the conclussion is simply that the type of sleep quality that ZQ is measuring is not what I'm looking for in terms of sleep quality, which would be to emphasize the ratio of restorative sleep, controlled over time, as a better measure of sleep quality.

My problem with the current ZQ is illustrated in a recent episode I had over the last weekend:  It takes a really, really bad night in terms of REM or Deep sleep in order to cancel out the positive score of just sleeping longer.  For examlpe, the other night I had food poisoning.  I slept very, very poorly (ick).  But I also knew I wasn't going to work the next morning, so I slept in.  My ZQ score was about the same as the night previously, when I'd slept orders of magnitude better, but rose early for work - because I slept a short amount one night well, and a long night the next, poorly, they sort of cancelled out.  Later, when I go back at the end of the month to see whether or not food poisioning is good or bad for sleep, my conclussion from the data I have from that one night is that food poisioning generally has no impact on ZQ score.

SOOOO!  All that said about ZQ score, it seems lazy of me to complain and wait fro someone else to do it for me, I've been working on some scripting tools over the weekend, sort of similar to the way that the Zeo Data Explorer works.  I've been defining my own variable, which I've been calling TruQ for now.

TruQ:

Here's how I'm calculating TruQ.  First off, my original proposal of just tracking percentage of sleep is also flawed, because that also changes with the amount of sleep you get.  The amount of time you spend in deep and REM sleep isn't the same early in the night compared to the evening, so the longer you sleep, the lower the REM and DEEP sleep will be in terms of total percentage of the night (you stop getting Deep sleep, for example).  So just using percentages does not adiquately control for sleep quality.

So to calculate a night's TruQ score, the script captures only the first 6 to 6.5 hours of sleep each night.  The start of night is defined as the first point in the night that you have two consecutive minutes of non-awake scores.  If the night does not have at least 6 hours of total records for the night, the night is discarded as not valid for data collection (I chose six hours because I almost always get at least six; others are probably different).  I drop any night with more than 5% of the total returned as missing data.

So, I'm looking at the percentage of sleep cycles over the first 6.5 hours only, every night that TruQ is calculated.  I then add REM and DEEP together as the "restorative" sleep, and divide them by the total minutes tracked by the system for the evening.  I include "awake" minutes in this, on the assumption that any average minute between start and finish that I'm awake is one I'd probably have rather been asleep, so is worth including in the total.  The resulting percentage is TruQ.

So far, my average TruQ score is about 41% on average.  My wife scores at about 48%, and she's an excellent sleeper, while I have sleep apnea and am a relatively poor sleeper.  I've talked a number of my friends into buying Zeo's so we can see if we can establish somewhat of a baseline.  Looking at the Zeo data, and pulling up the study that was used to measure change of sleep cycles over time, "average / healthy?" TruQ for my age is probably between 45 and 47%.

Now that I've got at least 30 days worth of data, I can start doing things like adding variables, and seeing if these things influnce this controlled variable.  I *think* this is a more reliable record of sleep quality than ZQ alone.  We'll see.  The script I wrote already extracts this information from the CSV files and stores the data for the six people I'm experimenting with (my friends that also have Zeos), but it just stores the data in a database at the moment.  I haven't built much of an interface for it, or I'd share graphs.  I'll let you know how it goes.

Other metrics I'm tracking:

As I mentioned, I have Sleep Apnea.  In comparing my sleep graphs to my wife's I notice that she gets all of her deep sleep in two or three long blocks up front at the start of the night.  I tend to get my deep sleep in short 3 to 6 minute blocks all throughout the night.  This makes sense, as sleep apnea events tend to happen the most in DEEP sleep, because your muscles relax the most there, and the throat closes.  I speculate that the result of this is that every time I'm reaching DEEP sleep, I relax, have an apena event, and get bumped back up to a lighter sleep.  Where I seem to come in lower than most people is that I am getting less DEEP than is "typical", as well as REM, to a lesser degree.

So I also programmed the tool to measure both the avareage number of minutes in length each Deep sleep block is per night, and how many individual blocks of deep sleep there are each night.  So far, this is turning out as a potential measure as well, as my wife has an average of 7 minutes of deep sleep in each block, compared to my 4 minutes.  Sadly, I don't have any other baseline than that, yet.  I wonder, though, if this is a measure of how sleep apnea can express itself in your sleep cycles.

Anyway, I'm having fun.  I'll keep you posted on these things as I get time.

Working with  Sleep Doctor:

A few days ago, I also happened to be discussing the Zeo and my observations with my sleep doctor, who is working with me with my sleep apnea.  He's agreed to set me up with a two night sleep study where we compare the results of some of their tools against the Zeo.  The purpose is not to validate the Zeo, but instead to determine whether or not the Zeo could be used in conjunction with the take home sleep studies, which don't have an easy way to tell if a patient is sleeping or not throughout the night; instead it measures apnea events, etc, but is thrown off if the gear keeps the person awake.

I'll let you know how that goes.

Best,

Aaron

I should check this forum more often.  I actually gave up on reporting the results on this forum a bit because my posts were taking two to three weeks to get approved, and at least one got eaten by the ether, I think.  Anyway, I'll post this update, and then see if it shows up, now that I see there's some kindred spirits out there.  :)

First, you might find this interesting.  After mentioning the Zeo to my sleep doctor, we started talking about the value of being able to easly judge the impact of sleep devices on sleep apnea.  Apparanly, most "take home" sleep studies will tell you when Apnea events occur, but don't measure if you are awake or asleep.  Consequently, if you're kept awake by the equipment, it shows up as no Apnea events during that time, etc.  So, testing the effectiveness of treatment is hard.

So, because he's a cool guy, he sent me home with a home sleep study for several nights just for fun, so we could compare the Zeo data with the same night from the sleep study.  The results are really quite interesting.  I'm pasting a picture below that is a composit overlay of both the Zeo data and the sleep study data from the same night.  I'll explain what you're seeing after the image:

Throughout the night, the Apnea events I experienced are the little black marks on the very first row of the image.  The Zeo data is overlayed on top of that, but upside down from how you'd normally see it displayed on the Zeo website, so that DEEP is represented as being at the top of the graph, and AWAKE is at the bottom.

What I think is interesting about this graphic is that I was looking for evidence of a specific hypothesis.  The hypothesis was that Apnea disproportiantely impacts DEEP sleep more than any other.  The logic behind this is that Obstructive Apnea events occur when the muscles in your throat relax the most.  Your body is never more relaxed than in DEEP and REM sleep, and so, my logic goes, every time you are in DEEP and REM sleep, you're more likely to have an Apnea event.  Second, that what an Apnea event does is basically wake you up for a second, bumping you back up to Light sleep.

As a result, someone with Apnea is expressed in the Zeo data as someone that gets substantially less restorative sleep than the average person (this is true for me).

So, if you look at the graphic above, I was really interested to see that this hypothesis is supported, I believe.  I've marked the main events in red, but you can see that every time I hit a DEEP sleep stage, I almost immediately have an Apnea event, and the event bumps me back up to Light sleep.  I think this is also the reason that sleeping aids make Apnea worse; yes you sleep more, but your body pushes you into Deep sleep more, as well, and you relax more, therefore more Apnea.

It's also worth complimenting the Zeo, as well, since it seems to line up with the data from the professional sleep study fairly well.

Update on TruQ:

Regarding the work I've been doing on TruQ, I've modified my opinions a little bit on what I'm trying to do.  Specifically, I have a slightly modified hypothesis/process I'm using to improve my sleep.  I'm beginning to believe that a proper sleep cycle - ideally - might be easiest to target by making sure that your body gets the right number of minimum hours of each type of sleep per week.  This has some relation to how the ZQ score works, and I'll connect it to the work on TruQ here in a second.

For example, if a healthy human 25 yr old adult male gets X number of total hours of cumulative DEEP sleep per week, and Y number of total hours of cumulative REM sleep per week, I consider this a reasonable target to aim for myself.  At the end of each week, if I've slept as much or more hours for each cycle as X and Y, and I think that this means my sleep will be comparable to a healthy sleeping cycle.  The sleep study showed that the impact of Apnea - in part, at least - is that it keeps you from getting blocks of DEEP and REM sleep as easily as another person, and so I'd expect to have to sleep more hours to achieve that.

This is where TruQ plays a subtastial role.  If you look at human sleep architecture, REM and DEEP are not distributed evenly throughout the night.  So while I can sleep longer and likely get more DEEP sleep, sleeping longer is unlikely to drive too much more REM sleep beyond a certain point.  So the question I find myself asking is, "Are there things that I can do to selectively increase the amount of a specific cycle I get each night?"

At the moment, my weekly average is that I get substantially less REM and DEEP sleep than the average person, but I can monitor this throughout the week.  If I know that I need to get more DEEP sleep tonight than anything else in order to meet my weekly quota of DEEP sleep, what factors can I modify to make this more likely?  Does meletonine help increase deep sleep?  Etc.  My intent is to use TruQ to help identify these fators, and then use the information to help target the outcome that I want.

My ultimate goal is to have a very clear roadmap of my sleeping patterns, so I can say, "On average, when condition A is true (exercised during the day, took meletonine, am wearing the CPAP), I have to sleep for an average of 7.4 hours per night to accumulate the same number of restorative hours as the average male of my age." vs, "when in condition B (exercised, but no CPAP machine), I have to sleep an average of 9 hours to get that same result."

The numbers above are made up at this point, but hopefully it illistrates my intent.

There was a question about how I'm putting together TruQ.  The answer is that I started with a spreadsheet, but quickly decided that it was a job for a website, so I put one together on my local computer.  It's pretty straight forward - it's intended to parse a CSV file downloaded from the Zeo website, and do the calculations for you for each night, and store the results for later comparissons.  It also stores variables for you, so that you can compare the average scores from different conditions.

Below is a quick screenshot of the front page of it, which is a summary comparing the TruQ scores of me and a number of my friends, as well, with their own Zeos.

To me, the table at the bottom is the most interesting.  For each person, it shows their TruQ score, and then the expected Average TruQ based on what I've been able to calculate from the sleep study and Zeo data I've gotten from the web, for each person's age group.  This is only limited in accuracy, because - while the TruQ scores are controlled for sleep time (6 to 6.5 hours) - I didn't have that same data for the "average" person.  If it helps, Maryam (my wife), Dan, and Brandon are all about the same age as I am (29 yrs), and none of them have sleep issues.  So, I use them as a base line.  I'm very jealus of Brandon, who beats out what I consider to be healthy by a large amount.

My father, Stewart, has severe sleep apnea, and is much larger.  His sleep is by far the worse.  This is with CPAP treatment, which he swears by.  I can't imagine what his scores looked like before that.  :)

Anyway, enough typing.  Let's see if this makes it through.  I'm writing this on April 22nd, 2012; curious to see when it gets approved.  If you'd like to e-mail me, my e-mail address is aaron@dreamwithme.net.  That's my pubic address, so I get a lot of spam - if you don't get a response, it's not that I'm ignoring you, but more likely I just never saw it come in.

Best,

Aaron

It posted right away!  Sweet - the settings on the forum must have been changed, because all my posts were being held for moderation review since my very first TruQ posting.  This will make like easier.

I'd edit my last post to use smaller images, but I don't seem to have permissions, so sorry about that.

Aaron